自适应动态能量特征提取的步态识别  

Gait Recognition with Adaptive Dynamic Energy Features Extraction

作　　者：聂栋栋[1] 马勤勇[2] 王毅[3] 

机构地区：[1]燕山大学理学院,河北秦皇岛066004 [2]燕山大学信息科学与工程学院,河北秦皇岛066004 [3]张家口供电公司,河北张家口075000

出　　处：《小型微型计算机系统》2014年第1期164-166,共3页Journal of Chinese Computer Systems

基　　金：河北省自然科学基金项目(A2011203053)资助;秦皇岛市科学技术研究与发展计划项目(2012021A044)资助

摘　　要：GEI算法对低质量的步态图像显示了较好的适应性,然而它更多的依靠人体的外形进行识别,因而在拍摄的人体外形变化较大时识别率明显降低.提出一个新的步态识别算法,以解决GEI的识别率受外形变化严重影响的问题.该算法首先根据左右基准点对步态轮廓图对齐.然后计算出每个关键时刻步态轮廓图相对于标准轮廓图的偏差,并根据这些偏差值生成动态能量矩阵.最后提取主分量并对特征矩阵进行分类.实验结果显示本文算法比GEI算法具有更高的识别率.这说明本文算法更能适应拍摄条件变化造成的轮廓图变形,并提取了更有效的步态动态特征.GEI algorithm shows good adaptability to the gait images of low quality, but it depends more on human body shapes to recognize identity, so it suffers low recognition rates to the image with big shape change. A new gait recognition algorithm is presented in this paper in order to solve the problem that the recognition rates of GEI are severely affected by the shape variations. In the new algorithm, silhouettes are firstly adjusted based on the left and the fight datum mark. Then the differences between each silhouette and standard silhouette are calculated by which a dynamic energy matrix is created. Finally, Principal component are extracted and utilized to classify objects. Experimental result shows that, compared with GEI algorithm, the proposed algorithm achieves higher recognition rate. It demonstrates this algorithm has better adaptability to silhouette deformation, and extracts more effective gait dynamic features.

关 键 词：生物特征 步态识别 特征提取 模式分类 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]